Describe the solution you'd like
Currently, we have our internal way of running multiprocessing jobs. Could we use something like DASK to make improve things? This would also mean that we could use MDANSE across multiple nodes on an HPC for instance.
DASK is specifically designed for computation with large arrays. This is quite suitable for our use case since we can have large MD simulations.
Additional context
Here is a screenshot from one of their YouTube video. In this example, they load up an HDF file and run some basic computations over it.
Describe the solution you'd like Currently, we have our internal way of running multiprocessing jobs. Could we use something like DASK to make improve things? This would also mean that we could use MDANSE across multiple nodes on an HPC for instance.
DASK is specifically designed for computation with large arrays. This is quite suitable for our use case since we can have large MD simulations.
Additional context Here is a screenshot from one of their YouTube video. In this example, they load up an HDF file and run some basic computations over it.